Analytic Development of Heterogeneous Patterns

• Activity Number: 506
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Hyun Kim*+ and Peiyun Lu and Eran Barnoy and David Gjertson
• Companies: University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles
• Keywords: heterogeneous patterns ; imaging ; visualization ; spatial information ; Approximate Profile-Likelihood Estimator ; asthma

Abstract

It is a challenge to describe and quantify the heterogeneous patterns within a subject or between subjects, and to rationalize the heterogeneous responses to a treatment in drug or device development. At the same time, it is an interesting developmental area in the personalized or tailored medicine. We explored the heterogeneity for analytical method in three steps: a) describe the pattern visually (usually high-dimension); b) quantify it with a heterogeneity index; c) connect medically identified heterogeneous patterns to a simple model. We used an example of respiratory disease, asthma. A descriptive summary statistics of eighteen spatial information is visualized with a plot of parenchymal structure. To quantify the patterns, we used index of Approximate Profile-Likelihood Estimator (APLE) with several simulations. Finally, we trained the visual categorical heterogeneous of air-trapping patterns at baseline and applied the classifier model to assess for changes over time to understand the intervention. The heterogeneous patterns and index can be a useful supportive tool for the enrichment of population or the illustrative rational in adapting designs.